Additional Latency Results?

[collinmccarthy]

Hi, do you have any additional latency results, maybe comparing your method to SimpleBEV, BEVFormer or FastBEV? I see Table 5 and Table 12 in the paper but I can't tell if this can be used as a real-time method for something like nuScenes (assuming INT8 quantization or similar).

Would you consider this to be a real-time method or not quite? If so do you have any additional latency results to share with respect to this?

Thanks.